CONFIDENTIAL. Limited circulation. For review only.
Screenbot: Walking Inverted Using Distributed Inward Gripping
Gregory D. Wile, Kathryn A. Daltorio, Eric D. Diller, Luther R. Palmer, Stanislav N. Gorb, Roy E.
Ritzmann, and Roger D. Quinn
Abstract—Insights from biology have helped reduce the
weight and increase the climbing ability of mobile robots.
This paper presents Screenbot, see Fig. 1, a new 126 gram
biologically-inspired robot that scales wire mesh substrates
using spines. Like insects, it walks with an alternating
tripod gait and maintains tension in opposing legs to keep
the feet attached to the substrate. A single motor drives all
six legs. Mechanisms were designed and tested to move the
spines into and out of contact with the screen. After the
spine engages the substrate, springs along the leg are
compressed. The opposing lateral spring forces constitute
a distributed inward grip that is similar to forces
measured on climbing insects and geckos. The distributed
inward gripping (DIG) holds the robot on the screen,
allowing it to climb vertically, walk inverted on a screen
ceiling and cling passively in these orientations.
G
I. INTRODUCTION
AIT patterns used by insects have often been used
in the design of fast, agile robots intended to
traverse irregular terrain. One common gait for
insect-inspired robots is the tripod gait. During this gait,
the front and rear legs on one side and the middle leg on
the other side are in stance, while the other three legs
swing together. For example, Prolero[1] and RHex[2]
both have simple rotating legs, but RHex’s tripod gait
results in improved fast walking. Whegs™ and MiniWhegs™ utilize an alternating gait as well, but each
rotating hub contains three or more spokes instead of
just one, allowing for smoother locomotion [3][4].
In addition to having an appropriate gait, climbing
animals must properly apply forces at each foot. In
quasi-static motion on a vertical surface, the feet must
support the body’s weight in shear and the front limbs
must provide a normal tensile force to counter the
moment produced by gravity about the center of mass.
Different animals support these tensile loads using
sticky pads, fine hair-like setae, claws, and spines [5].
Past robots built to climb vertical walls have utilized
suction [6][7], adhesion [8][9][10] and gripping
Manuscript received Feb 22, 2008. This work was supported by an
NDSEG Fellowship, an NSF Graduate Student Fellowship, AFOSR
under award number FA9550-07-1-0149, and by the Intelligence
Community (IC) Postdoctoral Fellowship Program under National
Geospatial Intelligence Agency contract HM1582-05-1-2021
G.D. Wile, K.A. Daltorio, E.D. Diller, L.R. Palmer, and R.D.
Quinn are with the Biologically Inspired Robotics Laboratory, Case
Western Reserve University, Cleveland, OH 44106 USA (phone: 216368-5216; e-mail: roger.quinn@case.edu) http://biorobots.case.edu
S. N. Gorb is with the Evolutionary Biomaterials Group at MaxPlanck- Institute for Metals Research, 70569 Stuttgart, Germany;
(email: S.Gorb@mf.mpg.de).
R. E. Ritzmann is with the Department of Biology at Case Western
Reserve University; (e-mail: roy.ritzmann@case.edu).
Fig. 1. Six-legged Screenbot walking on wire mesh.
handholds [11][12] to hold the body to the surface.
Later versions of Mini-Whegs™ have been modified to
climb smooth walls using gecko-inspired adhesive [13].
Spinybot uses an innovative array of compliant
microspines to climb hard, rough surfaces [14]. A 12degree-of-freedom robot, RiSE, uses spines to climb
curved surfaces such as trees [15]. Only a few robots
can walk inverted on ceilings. Some examples use
suction [16], vortex generating systems [17], and
adhesion [8].
For animals that adhere to surfaces, the motions and
forces during attachment and detachment are critical.
Geckos have feet covered with tenent (adhesive) setae
that conform to smooth and rough surfaces and adhere
with Van der Waals forces [18]. Flies also use tenent
setae to attach to a surface [19]. Setae on the feet of
animals are often angled away from the body, which
allows them to support a greater normal tensile adhesive
force when a tangential force is applied [5][18]. Flies
attach a foot by swinging it onto the substrate and then
sliding itt inward, toward the center of its body, until the
foot attaches. To detach its foot, the animal uses one of
four methods: shifting, pulling, twisting or rotating [19].
These motions may reduce the shear force and allow the
foot to detach more easily. Experiments with beetles
[20] on glass ceilings show that the beetles generally
attempt to distribute their contacting feet around the
center of mass (Fig. 2). For example, when all but two
adjacent ipsilateral feet are severed, the beetle quickly
falls from the ceiling. However, if the beetle has two
non-adjacent feet, it will stretch its legs to place the feet
so that they are directed opposite each other across the
center of mass, see Fig. 2E. In these configurations, the
animal can cling to the glass. From these experiments, it
appears that the animal is using the opposing tension of
Preprint submitted to 2008 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Received February 22, 2008.
CONFIDENTIAL. Limited circulation. For review only.
II. DESIGN
Fig. 2. Chrysolina fastuosa beetles clinging to a glass ceiling with
six feet (in A and B), three feet (in C and D) and 2 feet (in E and
F). The blue arrows indicate the orientation of the shear force
applied at the attached feet.
the legs to increase the shear at the foot of the animal so
that larger tensile normal forces can be supported. We
will refer to this principle of opposing tension of the
legs as “distributed inward gripping” (DIG). Because
the shear force is created only when opposing legs are
pulled inward, a beetle would be unable to remain
attached to a ceiling with only one foot in contact.
Cockroaches move their bodies laterally back and forth
while they walk and climb, suggesting that lateral forces
applied at their feet play a role in climbing performance
[21].
We hypothesized that these lateral forces would also
be valuable for climbing using hooks and spines. Spines
can be used to penetrate soft surfaces, cling to rough
substrates, or hook into holes on porous substrates.
Spine-climbing robots, like Spinybot on stucco walls
[14] or Climbing Mini-Whegs™ with hooks on steep
concrete [22], take advantage of gravity for the shear
force that holds the sharp hooks in contact with surface
asperities. Relying on DIG instead may allow climbing
on surfaces of any orientation or in zero-gravity because
the shear force is generated independently from the
weight of the robot.
In this paper, we test this hypothesis in the design of a
new robot, Screenbot, which uses these principles to
climb. Like the beetle, Screenbot relies upon mediallateral motions of its feet to grasp onto a surface. With a
sharp, angled spine on the tip of each foot, the robot can
climb and walk inverted on wire mesh screens.
Screenbot’s unique method of attaching to mesh walls
and ceilings may be valuable for future cleaning,
inspection and surveillance robots.
Each attachment mechanism described above is
optimized for some environments but have
disadvantages in others, which may explain why
animals often have multiple attachment mechanisms.
Suction and vortex based robots require a significant
amount of power, are inherently noisy, and rely upon
the presence of ambient pressure, excluding them from
use in some space applications. Magnetic end-effectors
only work on ferrous substrates. Adhesion-based robots
can cling to smooth surfaces, but can become
contaminated quickly by dust, dirt, or other substances
on the surface. Gecko-inspired structure may someday
solve the contamination problem in adhesives by
adhering even when dirty or by self-cleaning, but
strong, robust structured adhesives are still being
developed. Hooks and spines are able to grasp a wide
variety of naturally rough or porous substrates using
only frictional and compressive normal forces.
However, the application of spines requires precisely
timed force distributions and so far spine-bearing robots
have relied upon their weights to properly engage spines
and, thus, only work in specific climbing configurations.
Climbing Mini-Whegs™ is able to climb inclines of less
than 60° with pairs of sharp spines [22]. Spinybot [14]
can climb up vertically because it has many
independently compliant spines and a very low height to
length ratio, (see Section V). However, Spinybot cannot
climb in other orientations on the surface (like sideways
or forward-down).
There are several good reasons for using spines to
demonstrate DIG. Sharp rigid spines have directional
attachment properties based on well-understood friction
and interlocking rather than an adhesion model. The
concept of utilizing inward-pulling tangential force to
increase the attractive normal force on the feet can be
easily observed at a macroscopic scale using spines.
Also, spines require less frequent replacement than
many of the artificial insect-inspired directional
adhesives, which improves repeatability of our
experiments. Our design will not require the presence of
gravity. Instead, the “activating” force is produced
internally through the robot’s mechanisms via DIG. We
tested Screenbot on screens because the even mesh
provides an extremely uniform distribution of possible
footholds. In the future, other attachment mechanisms
that exhibit direction-dependent load bearing capacity,
such as peeled adhesives, anisotropic structured
adhesives, and arrays of hooks, could demonstrate DIG
on a similar platform in new environments.
For Screenbot to climb vertical surfaces as well as
walk inverted under horizontal surfaces, each foot must
be capable of supporting both normal and tangential
loads. When attached to a vertical substrate, the feet
above the center of mass must support tensile normal
forces to counteract the moment produced by the weight
Preprint submitted to 2008 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Received February 22, 2008.
CONFIDENTIAL. Limited circulation. For review only.
Fig. 3. The front right leg of Screenbot is shown in its uppermost
and lowermost positions. The bent aluminum axle (a) moves in a
circular path shown by the dashed arrows. When the spine (b) is
engaged with the screen, the spring compresses as the spine
retraction arm (c) pulls the foot inwards. The preload on the
spring is adjusted by moving (d). The spine retraction arm is
attached to the side of the chassis (e) with a nut and bolt. Its
horizontal base point is adjustable by sliding the connector (f).
the robot to fall or get stuck on the substrate. Therefore,
a mechanical control linkage can be designed to
automatically move the feet in the medial-lateral
c
direction at the correct times. As the proper timing was
not known during initial construction of the robot, the
timing mechanism was designed to be easily adjustable
by the user between runs. Also, the adjustable
mechanism allows the gait to be changed for walking on
various screen mesh sizes.
To minimize weight, the robot is driven with a single
drive motor. A tripod gait ensures that the feet are
synchronized. The three feet (the front and rear on one
side and the middle of the contralateral side) in stance
pull inward toward the centerline of the robot, while the
other three feet are in the swing phase, pushing outward
to detach.
The mechanical linkages (Fig 3.) make this specific
motion possible using only a single motor. Each leg of
the robot moves through the cycle depicted in Fig. 4.
The leg begins at the bottom of its cycle and moves
forward relative to the chassis (into the page). The leg
then begins to lower towards the screen (Fig. 4A) and
the spine engages the screen mesh. The retraction arm
then begins to “search” for a hold by pulling the spine
inwards (Fig. 4B) and connection is made with the
screen wire (Fig. 4C). The relative motion of the spine
retraction arm with respect to the foot will be termed
“foot compliance.” The stiffness of the compliance is
determined by the spring shown. The spring is
compressed after the spine engages the screen. The
weight of the robot will be supported by the sum of the
screen reaction forces shown in Fig. 4C. During the
final portion of the cycle, the retraction arm pushes the
foot outwards and the leg starts to pull away from the
mesh, reversing the process seen in Fig. 4. As the leg
detaches from the mesh, it moves down towards it
starting position. Note that all three legs in a tripod are
synchronized within this cycle, while the legs in the
other tripod follow the cycle out of phase by 180°.
III. CONSTRUCTION
Fig 4. The front right leg approaches the screen on its upstroke (A)
and begins to search for a hold as it pulls inward (B). When the
spine encounters a foothold, (a segment of mesh wire) the foot
complies to apply a ground reaction force at the mesh (C).
of the robot at some distance from the wall. Therefore,
the closer the center of mass is to the wall, or the longer
the robot is, the less adhesive force is required. Walking
inverted drastically changes these force requirements.
When the robot clings to the underside of a ceiling, the
tensile normal force at every foot must sum to
counteract the entire weight of the robot.
Another challenge is to enable the robot to release its
feet from the surface during walking. Furthermore, it
must not disengage a foot too soon or too late, causing
The Delrin chassis of the robot measures 127 x 38 x
22 mm, adding the legs extends the width to
approximately 140 mm. Spacing between adjacent legs,
where they attach to the chassis, is 45 mm. The weight
of the assembled robot is 126 grams.
The legs are made from 1/8-inch (3.18 mm) diameter
aluminum rod, bent at 15°. Pairs of legs are fit into
5/32-inch (3.97 mm) diameter brass tubes through the
chassis and are fixed 180° out of phase of each other
with small pins. The drive motor (model Hitec MX-52)
is connected to the three pairs of legs via a series of
Delrin chains and sprockets, depicted in Fig. 5. The
speed of the motor is radio-controlled and a 7.4 V
lithium polymer battery powers the vehicle.
Each leg consists of an assembly that slides freely
Preprint submitted to 2008 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Received February 22, 2008.
CONFIDENTIAL. Limited circulation. For review only.
Fig. 5. Underside of the robot, showing the drive system. The
drive motor is located behind the flat square panel near the top
of the picture.
along the straight outer portion of the aluminum rod. A
brass tube with a sliding fit on the aluminum rod serves
as the core of the assembly. A 25 mm long compression
spring with spring rate 80 N/m and maximum load of
1.4 N provides compliance to the sliding assembly,
which anchors the spine at the end of the leg. The
compression and release of this spring is controlled with
a spine retraction arm pinned to the chassis with a nut
and bolt. The horizontal position of the pin joint at the
chassis is adjustable. Delrin spine-bearing feet are pressfit to the sliding brass tube and are interchangeable for
easy replacement or modification. Each spine is a
sharpened brass pin.
IV. PERFORMANCE
The robot was tested on window screen made from
0.28 mm diameter aluminum wire and a square mesh
spacing of 1.5 mm. It was able to climb up vertically,
down a vertical surface, and inverted on a horizontal
screen “ceiling.” (See attached video.) For the initial
test, the spring stop (Fig. 3d) on each leg was adjusted
such that the spring was fully compressed, preventing
the foot from sliding relative to the spine retraction arm.
This was done to observe how the robot behaved with
zero foot compliance (completely rigid foot-leg
interface). As expected, the robot immediately stalled as
the feet attempted to pull the spines into the screen
further than the rigid configuration allowed. The
spring’s compressive length was then gradually
increased by moving the spring stop back until the robot
could take complete steps without stalling. This distance
of 5 mm was maintained for the remainder of the tests
performed.
Next, for each leg, the horizontal position of the spine
retraction arm relative to the center of the drive axle
(where it attached to the chassis) was adjusted (see Fig.
3f). The arm position, which we name “spine retraction
arm offset,” heavily influences the robot’s performance.
Initial testing was done with an offset of 0. Varying this
position from 0 to 12 mm forward of center, in
increments of 3 mm, the robot was run ten times for
each position at an average speed of 10 mm/sec, and an
average number of steps was recorded for each position.
This parameter changes the timing of the retraction and
extension of the foot as the leg rotates through its cycle.
As Fig. 6 shows, a greater offset results in a greater
average number of steps before falling. In four out of
ten trials with a 12 mm offset, the robot stalled. The
large offset delays the extension of the spine so much
that the legs try to pull out dorsally away from the
substrate before their respective spines are pushed
laterally
outward
a
sufficient
amount
for
disengagement. This behavior was observed only in the
rear legs and not the front or middle legs. This may be
because the center of mass is closer to the fore-legs, so
the tensile normal force due to gravity is greater in the
front, aiding the detachment. Further experimentation
found that the best performance occurred when the front
legs were set to an offset of 12 mm, the middle to 9 mm,
and the rear to 6 mm. This configuration resulted in 10
consecutive trials of at least 10 steps, and was used for
all following tests.
The next parameter tested was the angle of each spine
with respect to the axis of the leg. Previous tests were
done with spine angle of 45°. Running five trials at each
angle setting of 35°, 40°, 45°, 50° and 55°, the average
number of steps were again recorded for each. As Fig. 7
shows, the best performance is attained when the spine
angle is 45°. When the spines were less than 40°, the
robot was unable to lift them from the screen, resulting
in a stall. At greater than 50°, the spines consistently
disengaged prematurely, causing the robot to fall.
Further testing was done on a vertically oriented screen
to determine how the performance changed when the
direction of gravity was altered. When climbing
Fig. 6. The effect of the spine retraction arm offset distance
on inverted walking ability, measured as number of steps
before failure, with 10 trials averaged at each data point.
Fig. 7. The effect of spine angle on inverted walking ability
with 5 trials at each data point.
Preprint submitted to 2008 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Received February 22, 2008.
CONFIDENTIAL. Limited circulation. For review only.
Mass
TABLE I
CLIMBING ANIMALS FOOT FORCES
Cockroachesb
Fliesc
Geckosa
running
running
walking
vertically
vertically
inverted
3g
2g
0.1 g
Length
5 cm
5 cm
1.3 cm
Tensile
Normald
5 mN
7 mN
0.3 mN
Fore-Afte
20 mN
24 mN
Inward
Lateralf
0–13 mN
Normal/
Shear
Lateral/
Fore-aft
TABLE II
CLIMBING ROBOTS FOOT FORCES
Climbing MiniWhegs™a
Attaches
with:
Office
tape
Spinybotb
Stickybotc
Screenbotd
PVS
Directional
Single spines
Arrays of 20
structured
structured
spines
at foot
adhesive
adhesive
Mass
100 g
130 g
400 g
370 g
Length
7 cm
25 cm
58 cm
60 cm
1.3 mN
COM
Height
2 cm
2 cm
2 cm
3 cm
12-30 mN
Not measured
Tensile
~0.1 N
Normal
~0.05 N
~0.1 N
0.1 N
~0.3 N
22–26%
17–25%
At least 23%
Fore-Aft ~0.2 N
~0.3 N
~1.3 N
2.1 N
~0.4 N
0–64%
50–126%
~0 N
0.1–0.8 N
0.5–1.4 N
Measured forces at a single fore leg during stance for various
animals.
a
Gecko Hemidactylus garnotii data from [23]
b
Cockroach Blaberus discoidalis data from [21]
c
Blowfly Calliphora vicina data from [20], note that because fly
is walking on the ceiling, the fore-aft force is not in the direction of
the weight but due to opposing distributed inward gripping in the
fore-aft direction. Lateral forces were not measured in this
experiment.
d
Forces normal to the substrate supported by the adhesive
attachment mechanism
e
On a vertically climbing animal, these forces are directed up to
keep the animal from slipping down the substrate.
f
Forces that pull the animal laterally towards the same that the
foot is on.
vertically upwards, the robot was most successful when
the spine retraction arm offsets were as before, namely
12 mm, 9 mm, and 6 mm for the front, middle, and rear
legs, respectively. However, when the robot climbed
downwards, the rear legs tended to disengage
prematurely using this configuration. By changing the
offsets to 6 mm, 9 mm, and 12 mm for the front, middle,
and rear legs, respectively, downward climbing could be
achieved reliably. These findings support the idea that
for the uppermost legs that provide normal tensile force
with the screen, a greater offset is required to delay their
release as the opposite tripod begins to engage.
The length of individual spines was varied for vertical
126 g
8.7 cm
2 cm
Inward
Lateral
0N
0N
Normal/
Shear
57%
16%
10%
6%
19–45%
Lateral/
Fore-aft
0%
0%
0%
5–39%
114–345%
Approximated forces at a single fore leg during stance for some
recent robots with directional adhesives.
a
Climbing Mini-Whegs™ data from [13], approximations
calculated by assuming planar quasi-static system, weight
distributed evenly between four feet. The directional properties
come from the peeling of the adhesive.
b
SpinyBot data [14], approximate forces calculated assuming
planar quasi-static system, weight distributed between three feet in
stance, middle feet do not support adhesive normal force.
c
StickyBot forces measured in [9].
d
Screenbot data normal and fore-aft forces approximated by
assuming a quasi-static system with the weight distributed between
three feet in stance, middle feet do not support adhesive force.
Lateral forces are determined by observing spring deflection during
vertical climbing (0.9–1.8 cm) and multiplying by spring stiffness
(50 N/m).
climbing. The diameter of the spines used was .53mm
and the length of the spine in previous tests was 6.4mm.
The length of spine was measured from where the spine
is connected to the foot to the tip of the spine. Ten trials
of vertical climbing were conducted for each spine
length and the number of successful steps for each trial
was recorded. The failure mode for the last
(unsuccessful) step was also recorded. The first type of
failure is a binding failure. In this failure, the spines are
unable to release properly from the screen. This causes
the drive system to bind and the robot to stop forward
motion. As shown in Fig. 8, excessively long spines
result in increased binding. The second failure mode
occurs when one of the spines fails to engage the screen,
resulting in the robot falling away from the screen.
Shorter spine length increases the chance of a failed
engagement. A spine length between 7mm and 8mm
minimizes both types of failure.
V. COMPARISON WITH OTHER CLIMBERS
Fig. 8. The effect of spine length on the percentage of steps that
result in the robot binding or falling. Ten trials at each spine length.
To compare this robot with biological climbers and
existing robots, we can consider the forces at each foot.
To prevent the robot tumbling backwards from the
substrate, the attachment at the front feet is the most
Preprint submitted to 2008 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Received February 22, 2008.
CONFIDENTIAL. Limited circulation. For review only.
important. Table I compares the forces at the front legs
for two well-documented vertical-climbing animals:
cockroaches and geckos. Both of these animals push
outward with their legs when walking on the flat ground
but switch to pulling inward (DIG) on vertical walls
[21][23]. For these animals, as well as for blowflies
[20], the normal forces are about one quarter of the
tangential (shear) foot forces during stance. The shear
forces have significant components in the lateral
direction and in the fore-aft direction. When the animal
is climbing vertically, the fore-aft forces at each foot
must sum to equal the weight plus any upward
acceleration. Table II shows that recent robots have
made increasingly effective use of fore-aft shear forces
to sustain attachment. Long, close-to-the-wall robots
leverage relatively weak attachment devices like
structured PVS or arrays of hooks to cling to vertical
surfaces. However, of these robots, Screenbot is the first
to take advantage of lateral shear forces of nearly the
magnitude of the fore-aft shear forces.
For this reason, Screenbot is the only one of the
robots in Table II, to climb inverted on ceilings as well
as vertical surfaces using biologically-inspired adhesive.
(Climbing Mini-Whegs™ with office tape can walk on
ceilings for short periods because office tape was
adhesive enough even with little shear [8].) On the
ceiling, gravity loads each foot with tensile normal
forces, and the shear forces must be provided by a
distributed grip of opposing forces, in this case inward
forces or DIG.
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
VI. CONCLUSIONS
Screenbot is the first robot to our knowledge that can
climb inverted using only spines. Future versions may
include turning capability or even a body joint to aid in
transitions between surfaces. While the parameters of
the robot may be adjusted to climb on substrates with a
less even distribution of footholds, one of the major
limitations of Screenbot is that it requires a rough
surface that contains pores or asperities.
Screenbot demonstrates the effectiveness of DIG on
rough surfaces, but the data in Tables I and II indicate
that significant shear forces are important for many
different types of successful climbers with many
different kinds of biological and biologically-inspired
adhesives. Therefore generating opposing forces in the
lateral direction as in this work, and in the fore-aft
direction as in animals may be important for securing
many types of future surface-walking robots whether
they are on smooth or rough surfaces.
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
REFERENCES
[1]
A. Martin-Alvarz, W. de Peuter, J. Hellebrand, P. Putz, A.
Matthyssen, and J. R. de Weerd, “Walking Robots for Planetary
Exploration Missions,” in Proc. of 2nd World Automation
Congress, Montpellier, France, May, 1996.
[23]
U. Saranli, M. Buehler, and D. Koditschek, “RHex: a simple and
highly mobile hexapod robot,” International Journal of Robotics
Research, vol. 20, no. 7, pp. 616-631, 2001.
T. J. Allen, R. D. Quinn, R. J. Bachmann, and R. E. Ritzmann,
“Abstracted biological principles applied with reduced actuation
improve mobility of legged vehicles,” in Proc. of Int. Conf. on
Intelligent Robots and Systems (IROS), Las Vegas, Nevada,
2003.
J.
M.
Morrey,
B. Lambrecht, A.
D. Horchler, R.
E. Ritzmann, and R. D. Quinn, “Highly mobile and robust small
quadruped robots,” in Proc. of Int. Conf. on Intelligent Robots
and Systems (IROS), Las Vegas, Nevada, 2003.
S. N. Gorb, Attachment devices of insect cuticle, Dordrecht,
Boston, London: Kluwer Academic Publishers, pp. 1–305, 2001.
S. Hirose, A. Nagabuko, and R. Toyama, “Machine that can
walk and climb on floors, walls, and ceilings,” in Proc. of ICAR
‘91, pp. 753-758, Pisa, Italy, 1991.
T. Yano, T. Suwa, M. Murakami, and T. Yamamoto,
“Development of a semi self-contained wall climbing robot with
scanning type suction cups,” in Proc. of IROS ’97, Grenoble,
France, 1997.
K. A. Daltorio, A. D. Horchler, S. Gorb, R. E. Ritzmann, and R.
D. Quinn, “A small wall-walking robot with compliant, adhesive
feet,” in Proc. of IROS ’05, pp. 3648-3653, August, 2005.
S. Kim, M. Spenko, S. Trujillo, B. Heyneman, V. Mattoli, M. R.
Cutkosky, “Whole body adhesion: hierarchical, directional and
distributed control of adhesive forces for a climbing robot,” in
Proc. of ICRA ’07, pp. 1268-1273, Roma, Italy, April, 2007.
C. Menon and M. Sitti, “Biologically-inspired adhesion based
surface climbing robots,” in Proc. of ICRA ’05, pp. 2715-2720,
2005.
T. Bretl, S. Rock, J. C. Latombe, B. Kennedy, and H.
Aghazarian, “Free-climbing with a multi-use robot,” in Proc. of
Inter. Symp. on Experimental Robotics (ISER), Singapore, June,
2004.
H. Amano, K. Osuka, and T. J. Tarn, “Development of vertically
moving robot with gripping handrails for firefighting,” in Proc.
of IROS ‘01, Maui, HI, 2001.
K. A. Daltorio, S. Gorb, A. Peressadko, A. D. Horchler, R. E.
Ritzmann, and R. D. Quinn, “A robot that climbs walls using
micro-structured polymer feet.” in Proc. of CLAWAR ’05,
London, UK, September, 2005.
S. Kim, A. T. Asbeck, M. R. Cutkosky, and W. R. Provancher,
“SpinybotII: climbing hard walls with compliant microspines,”
in Proc. of IEEE Int. Conf. on Advanced Robotics (ICAR ‘05),
Seattle, 2005.
K. Autumn, M. Buehler, M. Cutkosky, R. Fearing, R.J. Full, D.
Goldman, R. Groff, W. Provancher, A. A. Rizzi, U. Saranli, A.
Saunders, and D. E. Koditschek, “Robots in scansorial
environments,” in Proc. of SPIE, vol. 5804, pp. 291–302, 2005.
S. Hirose and K. Kawabe, “Ceiling walk of quadruped wall
climbing robot NINJA-II,” in Proc. of CLAWAR ‘98, Brussels,
Belgium, 1998.
J. Xiao, W. Morris, N. Chakravarthy, A. Calle, “City climber: a
new generation of mobile robot with wall-climbing capability,”
in Proc. of SPIE. vol. 6230, pp. 62301d, 2006.
K. Autumn, Y.A. Liang, S.T. Hsieh, et al. “Adhesive force of a
single gecko foot-hair,” Nature, vol. 405, pp. 681–685. 2000
S. Niederegger and S. Gorb, “Tarsal movements in flies during
leg attachment and detachment on a smooth substrate,” Journal
of Insect Physiology, vol. 49, pp. 611-620, 2003.
S.N. Gorb, personal communication, unpublished data.
Goldman, D., et al. “Dynamics of rapid vertical climbing in
cockroaches reveals a template”. Journal of Experimental
Biology, vol. 209. 2006.
T.E. Wei, K.A. Daltorio, S.N. Gorb, L. Southard, R.E.
Ritzmann, and R.D. Quinn “A Small Climbing Robot with
Compliant Ankles and Multiple Attachment Mechanisms,” Int.
Conf. on Climbing and Walking Robots (CLAWAR ’06),
Brussels, Belgium. 2006.
Autumn, K. et al. “Dynamics of geckos running vertically,”
Journal of Experimental Biology vol. 209, pp. 260-272. 2006.
Preprint submitted to 2008 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Received February 22, 2008.